WDR5 — the partner that is not a partner: it binds the half of KMT2A the fusion throws away

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• Essential-function toxicity is the headline risk: WDR5 is DepMap common-essential with a mean Chronos of -1.159 and 92.6% of lines depleted. There is no therapeutic window to optimise into.
• The interface being targeted is absent from the disease protein. A WIN-site inhibitor in KMT2A-rearranged leukemia is acting on wild-type KMT2A, KMT2B, SETD1A and other WIN-motif proteins, not on the fusion.
• Excellent chemistry is a trap here, not an asset: 1,662 potent data points and a 1 nM compound make the target look de-risked when the structural and dependency logic both say it is mis-specified for this disease.
• WDR5 is a hub in multiple complexes (WRAD, NSL/KANSL, MYC). Inhibiting it perturbs all of them, and the sampled chemical matter does not distinguish which site a given compound occupies.
• If the project cites WDR5 as a validated KMT2A-complex target in a grant or a paper without the residue mapping, a reviewer who checks 3EG6 will find residue 3765 and the argument collapses. This node exists so that does not happen.

1. Does WDR5 inhibition act in KMT2A-rearranged cells through the remaining wild-type KMT2A allele? KMT2A itself is the second-most selective dependency in the ranking, so wild-type KMT2A function may matter in these cells and WDR5 inhibition would hit that rather than the fusion. Comparing a WIN-site inhibitor in KMT2A-wild-type-allele-deleted versus intact KMT2A-rearranged cells would separate the two.
2. Is the anti-selectivity a real biological signal or an artefact of the 10-line sample? The monotonic gradient across four nested contexts argues it is real, but a leave-one-out test has not been run.
3. Do any KMT2A breakpoints retain enough C-terminal sequence to preserve the WIN motif? Across 107 in-frame fusions in the recombinome catalogue this should be checkable, and if any do, they would define a subgroup for which WDR5 genuinely is a fusion partner.
4. Does the MYC-binding site on WDR5, as opposed to the WIN site, have any KMT2A-rearrangement-selective relevance? Much of the 1,662-row chemical effort is aimed there and this node's argument does not cover it.
5. Why is there no gnomAD constraint record for WDR5? A 334-residue gene with mis_z 4.80 should have one, and the gap should be resolved before any tolerability argument is built on constraint data.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Every structure of WDR5 or the WRAD module with KMT2A models only C-terminal KMT2A residues: 3EG6 (WIN motif, KMT2A 3762-3773, 1.72 A), 3EMH (3764-3776, 1.37 A), 3P4F (3761-3770 with RBBP5 371-381, 2.35 A), 6KIU (3754-3969 with WDR5, RBBP5, ASH2L and a ubiquitinated nucleosome, 3.20 A), 6PWV (3762-3969, full core complex on the nucleosome, 6.20 A) and 7W67 (3813-3969 with RBBP5 330-356 and ASH2L, 2.19 A). WDR5 also has WIN-motif complexes with KMT2B (3UVK, 1.40 A) and SETD1A (3UVN 1.79 A, 4EWR 1.50 A) and numerous histone H3K4me peptide complexes. No structure of WDR5 with any KMT2A fusion protein exists.

   RCSB PDB entries 3EG6, 3EMH, 3P4F, 6KIU, 6KIV, 6KIX, 6KIZ, 6PWV, 6PWW, 6W5I, 7MBM, 7W67, 3UVK, 3UVN, 4EWR and the WDR5-histone series, with per-entity UniProt accessions and modelled residue ranges resolved against the RCSB search and data GraphQL APIs · RCSB Protein Data Bank · RCSB PDB 2026
   ▸What this evidence cannot show

   Residue ranges are the modelled or deposited construct ranges as RCSB annotates them against UniProt Q03164, not independently curated motif boundaries. Absence of a WDR5-fusion structure is absence of evidence, and could reflect that nobody has attempted it rather than that no interaction exists - although the residue mapping supplies a mechanistic reason.

   ▸Evidence that argues against this (1)

   • Every structure modelling only C-terminal KMT2A residues is the decisive fact for a fusion dossier: the WIN motif at 3762-3776 is LOST in every KMT2A fusion, since breaks cluster between intron 7 and exon 13. The structures describe an interaction the oncoprotein cannot make. Leukemia 2023
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The 1.7 A crystal structure of WDR5 bound to the MLL1/KMT2A WDR5-interaction (WIN) motif peptide identifies Arg-3765 of MLL1 as the arginine occupying the WDR5 pocket, and shows the WIN peptide is preferentially recognised over histone H3.

   Structure of WDR5 bound to mixed lineage leukemia protein-1 peptide · Patel A, Dharmarajan V, Cosgrove MS · Journal of Biological Chemistry 2008
   PMID 18829459DOI
   ▸What this evidence cannot show

   Thermodynamic binding measurements were performed in this work but no numeric Kd appears in the abstract, so no affinity is attributed here. The structure is an isolated peptide with an isolated domain.

   ▸Evidence that argues against this (1)

   • Arg-3765 occupying the WDR5 pocket is wild-type KMT2A biology. The fusion retains the N-terminus, so a WIN-motif inhibitor acts on the remaining wild-type allele rather than on the oncoprotein — which may still be therapeutic, and is a different mechanism from the one the entry implies. Cell Reports 2014
3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The KMT2A recombinome comprises genomic breakpoints within KMT2A across 3,401 patients and 107 in-frame KMT2A fusions, establishing that breakpoints cluster in the 5' region of the gene and that fusion proteins retain the KMT2A N-terminus while losing C-terminal domains.

   The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023
   PMID 37019990DOI
   ▸What this evidence cannot show

   The abstract does not state a residue count for what fusions retain, so the specific claim that the WIN motif at residue 3765 is excluded is our inference from the breakpoint architecture plus the residue mapping in E1 and E2, not a statement in this paper. Breakpoint heterogeneity is substantial and a universal claim would be overreaching; the argument applies to the common intron 9-11 breakpoints.

   ▸Evidence that argues against this (1)

   • The recombinome is cited here for what the fusion retains, and this graph's own audit finds it supporting 75 items across 69 entries in 14 beats, graded at four different confidence levels. Its breakpoint percentages are also detection-limited, since a census counts the rearrangements the contributing laboratories could characterise. Oncofusion knowledge graph (internal analysis) 2026
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   WDR5 is DepMap common-essential with a mean Chronos gene effect of -1.159 across 1,258 cell lines, yet is depleted less in KMT2A-rearranged lines (-1.051) and less again in KMT2A::AFF1 lines (-0.868) than in leukemia lines generally (-1.242), giving a lineage-controlled selectivity of +0.191 - the worst of 41 candidate partners. gnomAD returned no pLI or LOEUF record for WDR5 in this query; its missense Z-score is 4.80.

   DepMap CRISPR (Chronos) gene effect and essentiality flags via the Open Targets Platform GraphQL API; loss-of-function constraint via the gnomAD GraphQL API; normal-tissue expression via the Human Protein Atlas API · Broad Institute DepMap; gnomAD; Human Protein Atlas; Open Targets Platform · Databases 2026
   ▸What this evidence cannot show

   Means were computed by us from per-line API values, not read from a DepMap release table, and the release version is not stated by the API. Only 10 KMT2A-rearranged lines (4 KMT2A::AFF1) contribute, so +0.191 carries wide uncertainty; the claim is strengthened by being monotonic across four nested contexts rather than by the point estimate alone. The absent gnomAD record is a database gap and must not be read as absence of constraint.

   ▸Evidence that argues against this (1)

   • WDR5 remains a deep dependency in absolute terms in KMT2A-rearranged lines (-1.051), so WDR5 inhibitors do kill these cells. The argument here is about therapeutic index and about which KMT2A residues the interaction uses, not about activity. Broad Institute DepMap; Open Targets Platform 2026
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   ChEMBL v34 target CHEMBL1075317 (WD repeat-containing protein 5) carries 1,662 activity rows at pChEMBL >= 6 with a best IC50 of 1 nM (CHEMBL1082290), the largest chemical-matter footprint among the KMT2A-complex proteins, and has no clinical candidate with a curated mechanism of action.

   ChEMBL database version 34, target CHEMBL1075317 (WD repeat-containing protein 5): target_search, get_bioactivity and get_mechanism queries · EMBL-EBI ChEMBL · ChEMBL 2026
   ▸What this evidence cannot show

   The sampled page of 25 rows was entirely MLL1 WIN-motif peptides and fluorescence-polarisation tracers rather than small molecules, so the best IC50 of 1 nM may belong to a peptide, and the small-molecule chemotypes sit in the roughly 1,637 unsampled rows. The rows do not distinguish WIN-site from MYC-site binders, and much of the WDR5 medicinal-chemistry effort is directed at the MYC-binding site, to which this node's argument does not apply.

   ▸Evidence that argues against this (1)

   • A ChEMBL activity-row count measures how much medicinal chemistry has been DONE, not how tractable the target is, and the best reported affinity belongs to the most optimised compound in whichever series someone chose to pursue. Zero rows means nobody tried; thousands of rows means a target had a decade of industrial attention. 1,662 rows is the largest footprint in this interactome and reflects that WDR5 became a fashionable target through MYC rather than through KMT2A. A 1 nM IC50 against a WIN-site binder also acts at a site the fusion has lost. Leukemia 2023
6. E_yokoyama2004strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Biochemical purification of MLL identified a SET1-like complex containing ASH2L, WDR5, RBBP5, HCF-1, HCF-2 and menin. Oncogenic mutant (fusion) forms of MLL retain the ability to interact with menin but not the other identified complex components.

   Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression · Yokoyama A, Wang Z, Wysocka J, Sanyal M, Aufiero DJ, Kitabayashi I, Herr W, Cleary ML · Molecular and Cellular Biology 24:5639-49 2004
   PMID 15199122DOI
   ▸What this evidence cannot show

   Read as abstract only by the coordinator; the interactome beat reports reading the full text via PMC480881. 2004 affinity-purification mass spectrometry, at depths well below current practice, so a negative here is a failure to detect rather than a demonstration of absence. No proximity-labelling study of any KMT2A fusion has ever been published, so this negative has never been re-tested with a method capable of catching transient or substoichiometric contacts. **The founding negative is an inference from a deletion mutant, not a measured fusion co-immunoprecipitation.** Read in full 2026-09-20: WRAD subunits were mapped using a C-terminal deletion construct of otherwise wild-type MLL in 293 cells. The paper's only fusion experiments blot for HIS, MLL-C, HCF-1, menin, MLL-N and ENL — **no WRAD subunit is blotted on any fusion anywhere in it**. It is a measured negative for HCF-1 only.

   ▸Evidence that argues against this (4)

   • A 2004 co-purification negative is weak evidence of non-interaction. Modern proximity labelling routinely recovers contacts that affinity purification misses, particularly transient ones, and the entire composition literature for this fusion rests on stable-complex purifications from 2004-2012. The claim here should be read as 'not detected as a stable complex member', not 'does not interact'. Yokoyama A et al. 2004
   • CORRECTED 2026-09-21 (cycle 44), and the previous version of this note was too strong. It said "zero published experiments test WRAD-versus-KMT2A-fusion association in either direction" and cited PMID 15199122 for that - a 2004 paper, which cannot establish an absence as of today, in a note whose own next sentence warned against citing that paper for more than it shows. Re-bounded by query today: the subunit-by-fusion co-occurrence search is NOT empty (WDR5 26 records, ASH2L 6, DPY30 3, RBBP5 2), and reading all 26 WDR5 titles returns two that genuinely bear on the question. PMID 27462455 shows MLL1 and MLL-AF9 occupy DISTINCT chromatin regions and attributes MLL1's enhancer recruitment to a C-terminal WDR5 interaction - hedged as "probably" by its own authors - which is an experiment discriminating wild-type from fusion with WDR5 as the proposed discriminator. PMID 36402263 shows genetic Wdr5 deletion attenuates MLL-AF9 leukemogenesis, acting through ribosomal genes rather than through fusion binding. THE SURVIVING ABSENCE IS NARROWER AND STILL WORTH STATING: no direct binding assay - co-IP, structure or proximity labelling - has tested whether WRAD associates with a KMT2A FUSION protein. The two experiments above are genomic occupancy and genetic requirement, neither of which is a binding measurement. The structural argument for exclusion is unchanged, because the WRAD contacts sit thousands of residues past the breakpoint. What this graph must not do is cite a 2004 deletion-mutant result as if it were a fusion co-IP negative. PubMed (bounded query, counts and date recorded) 2026
   • The nearest thing to a published test, and it cuts against the absence being total: MLL1 and MLL-AF9 occupy distinct chromatin regions, and the authors attribute MLL1's enhancer recruitment to a C-terminal WDR5 interaction that the fusion lacks. It is an occupancy experiment, not a binding assay, and the WDR5 attribution is hedged in the source. Xu J et al. 2016
   • And a functional requirement in the fusion context: genetic Wdr5 deletion attenuates MLL-AF9 leukemogenesis in vivo. That WDR5 is REQUIRED where the fusion drives disease does not show WRAD binds the fusion - the reported mechanism is ribosomal gene regulation - but it makes 'WRAD is irrelevant to the fusion' untenable. Liu L et al. 2023
7. E_thielstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Menin recruits both wild-type MLL and the oncogenic MLL-AF9 fusion protein to HOX gene loci to activate transcription, and the wild-type Mll allele is required for MLL-AF9-induced leukemogenesis and for maintenance of MLL-AF9-transformed cells.

   MLL-AF9-induced leukemogenesis requires coexpression of the wild-type Mll allele · Thiel AT, Blessington P, Zou T, Feather D, Wu X, Yan J, Zhang H, Liu Z, Ernst P, Koretzky GA, Hua X · Cancer Cell 17:148-59 2010
   PMID 20159607DOI
   ▸What this evidence cannot show

   Read as abstract only. Mouse genetics with a retroviral MLL-AF9 model; the requirement is shown for MLL-AF9, not for KMT2A::AFF1, and this graph has repeatedly found partner-specific differences. 'Required' here is allele-level and does not identify which wild-type MLL module carries the requirement.

   ▸Evidence that argues against this (1)

   • A wild-type-allele requirement shown for MLL-AF9 in mouse need not hold for KMT2A::AFF1 in infant B-ALL, and this project has already been burned generalising across partners. The claim should be carried as strong for MLL-AF9 and untested for the fusion this project is about. Thiel AT et al. 2010
8. E_milnestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The essential MLL1 recruitment domain includes both the CXXC domain and the PHD fingers and is controlled by interactions with the PAF elongation complex and H3K4me2/3. MLL1 fusion proteins lack the PHD fingers and require prebinding of a wild-type MLL1 complex, together with CXXC recognition of DNA, for stable HoxA9 association.

   Multiple interactions recruit MLL1 and MLL1 fusion proteins to the HOXA9 locus in leukemogenesis · Milne TA, Kim J, Wang GG, Stadler SC, Basrur V, Whitcomb SJ, Wang Z, Ruthenburg AJ, Elenitoba-Johnson KSJ, Roeder RG, Allis CD · Molecular Cell 38:853-63 2010
   PMID 20541448DOI
   ▸What this evidence cannot show

   Read as abstract only. Establishes a trans requirement at HoxA9 specifically; whether it generalises to the whole fusion target set is not addressed in the abstract. The authors themselves frame the distinct recruitment mechanisms as a therapeutic opportunity rather than an obstacle.

   ▸Evidence that argues against this (1)

   • If the fusion needs the wild-type complex prebound, then wild-type MLL1 recruitment is a potential target — but it is also a normal function in every cell that expresses MLL1, so the therapeutic window argument that applies to menin does not automatically transfer here. Milne TA et al. 2010
9. E_caostrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   MM-401 inhibits MLL1 activity by blocking the MLL1-WDR5 interaction and preventing complex assembly, without affecting other MLL-family methyltransferases, and specifically blocks proliferation of MLL cells by inducing cell-cycle arrest, apoptosis and myeloid differentiation without general toxicity to normal bone marrow cells or non-MLL cells.

   Targeting MLL1 H3K4 methyltransferase activity in mixed-lineage leukemia · Cao F, Townsend EC, Karatas H, Xu J, Li L, Lee S, Liu L, Chen Y, et al. (Dou Y, senior author) · Molecular Cell 53:247-61 2014
   PMID 24389101DOI
   ▸What this evidence cannot show

   Read as abstract only. A peptidomimetic tool compound characterised in cells; no clinical or in vivo efficacy is claimed in the abstract, and 'MLL cells' spans MLL-rearranged lines generally rather than KMT2A::AFF1 specifically. SUPERSEDED IN PART (2026-09-19): MM-401's stated mechanism is inhibition of MLL1 methyltransferase activity, but WIN-site inhibition in MLL-rearranged cells is reported to act through ribosome protein gene suppression and p53 activation (PMIDs 30865883, 38682900). This citation should no longer be used as evidence that a wild-type-only MLL1 interface is druggable through MLL1 biology.

   ▸Evidence that argues against this (1)

   • MM-401's stated mechanism is inhibition of MLL1 H3K4 methyltransferase activity, yet the MLL1 SET domain is reported dispensable for maintaining HSCs and for MLL-AF9-driven leukemogenesis (Mishra 2014, PMID 24813891), which this graph also carries. Either the selectivity of MM-401 runs through something other than catalysis, or the two results are in tension. The graph should hold both rather than choose. Mishra BP et al. 2014
10. E_ahostrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Potent WIN-site inhibitors show that the WIN site links WDR5 to chromatin at a small cohort of loci including a specific subset of ribosome protein genes; the inhibitors rapidly displace WDR5 from chromatin and decrease expression of those genes, causing translational inhibition, nucleolar stress and p53 induction.

    Displacement of WDR5 from Chromatin by a WIN Site Inhibitor with Picomolar Affinity · Aho ER, Wang J, Gogliotti RD, Howard GC, Phan J, Acharya P, et al. (Tansey WP, senior author) · Cell Reports 26:2916-2928.e13 2019
    PMID 30865883DOI
    ▸What this evidence cannot show

    Read as abstract only. Establishes the mechanism of small-molecule WIN-site inhibitors across cancer contexts; it is not specific to KMT2A::AFF1 and does not directly test the peptidomimetic MM-401.

    ▸Evidence that argues against this (1)

    • MM-401 and the small-molecule WIN-site inhibitors are different chemotypes, and this paper does not test MM-401. It remains formally possible that the peptidomimetic acts differently. But the burden has shifted: the mechanism cannot now be assumed to be MLL1-programme disruption without direct evidence, and this graph previously assumed exactly that. Howard GC et al. 2024
11. E_howardstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MLL-rearranged cancer cells, WDR5 WIN-site inhibition acts by suppressing ribosome protein gene transcription, inducing nucleolar stress and activating p53, with depletion of the ribosome stock, a broad modest translational choke, and alternative-splicing changes that inactivate the p53 antagonist MDM4. The authors state that peptidomimetic WIN-site inhibitors were originally proposed to act via genes connected to haematopoietic stem cell expansion, and that their data show otherwise. WIN-site inhibitors are synergistic with venetoclax and BET-bromodomain inhibitors.

    Ribosome subunit attrition and activation of the p53-MDM4 axis dominate the response of MLL-rearranged cancer cells to WDR5 WIN site inhibition · Howard GC, Wang J, Rose KL, Jones C, Patel P, Tsui T, et al. (Tansey WP, senior author) · eLife 12:e90683 2024
    PMID 38682900DOI
    ▸What this evidence cannot show

    Read as abstract only. Cell-line work in MLL-rearranged lines; the p53 dependence is inferred from the mechanism rather than demonstrated across TP53-mutant and wild-type isogenic pairs in the abstract. Does not address KMT2A::AFF1 infant B-ALL specifically.

    ▸Evidence that argues against this (1)

    • A p53/ribosome mechanism does not make WIN-site inhibition useless in MLL-rearranged leukaemia - the same paper reports synergy with venetoclax and BET inhibitors, and a ribosome-directed strategy is novel in its own right. What it removes is the specific inference this graph drew: that MM-401 demonstrates an absent-from-the-fusion interface can be drugged through MLL1 biology. Howard GC et al. 2024

• ← related toHypothesis: transfer-RNA and ribosome biology is a genuinely empty cell in KMT2A-rearranged leukemia, and translational capacity may be the unmeasured dependency

  2026-09-19: the 'empty cell' claim withdrawn - WDR5 WIN-site inhibition is a measured ribosome/nucleolar/translation phenotype in MLLr cells (PMID 38682900)

• ← part ofWDR5 maintains a differentiation block — in CEBPA p30 AML, not in KMT2A-rearranged disease, and a title check nearly buried it

  the cell-context dossier for this gene

• regulates →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia

  required to maintain the myeloid differentiation block in murine CEBPA p30 AML - NOT shown in KMT2A-rearranged disease

• regulates →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell

  the p30-driven increase in self-renewal requires Wdr5 (murine CEBPA p30 AML)

• ← related toHypothesis — the menin lesson generalises: every non-enzymatic fusion has an obligate partner interface that is more druggable than the fusion

  Counter-evidence to the hypothesis's central inference (Aho 2019 PMID 30865883; Howard 2024 PMID 38682900): the WDR5 WIN site satisfies every structural criterion this hypothesis lists, and its picomolar inhibitors act in MLL-rearranged cells through ribosome protein gene suppression, nucleolar stress and p53/MDM4 -- not through the partner's chromatin biology. Interface engagement does not entail interface-mediated mechanism.

• ← related toAnalysis (planned) — comparative druggability scan of every obligate fusion-partner interface in the atlas

  Design flaw in the scan (Aho 2019 PMID 30865883; Howard 2024 PMID 38682900): the WDR5 WIN site is a structurally ideal, picomolar-liganded partner interface whose inhibitors act through ribosome depletion and p53/MDM4. Scoring pocket character plus ligand availability ranks it highly and scores the wrong mechanism; a separation-of-function column is required.

• ← related toWhy PML::RARA became curable — and what KMT2A::AFF1 does not share with it

  Degrade the partner, not the fusion (Yu 2021, PMID 34586829): a VHL-recruiting WDR5 degrader outperformed WIN-site occupancy inhibitors on MLL-complex chromatin binding, c-MYC and MLL-r AML patient cells in vivo. The degradation-resistance disanalogy assumes the substrate must be the fusion.

• ← related toExperiment (proposed) — cross-family interface point-mutant panel: is the obligate interface real in each fusion?

  Add a degradation arm (Yu 2021, PMID 34586829): genetic interface mutant, occupancy inhibitor and degrader can disagree, and the WDR5 case is one where occupancy failed and degradation worked.

• related to →Hypothesis: KMT2A::AFF1 rewires splicing, because AFF1's own protein family are G-quadruplex RNA-binding splicing modulators and the fusion keeps AFF1

  Confound counter (Howard 2024, PMID 38682900): WDR5 WIN-site inhibition alone produces alternative-splicing changes that inactivate MDM4 in MLL-rearranged cells - a drug-imposed splicing change in exactly this disease, showing that a splicing signature without a fusion-knockdown arm cannot be attributed to the fusion

• related to →Hypothesis: transfer-RNA and ribosome biology is a genuinely empty cell in KMT2A-rearranged leukemia, and translational capacity may be the unmeasured dependency

  Counter-evidence (Howard 2024, PMID 38682900): the one ribosome phenotype reported in MLL-rearranged cells is WDR5 WIN-site-inhibitor-imposed - ribosome protein gene suppression, nucleolar stress, p53 activation, MDM4-inactivating splicing - and it replaced the authors' earlier MLL1-biology explanation; ribosome biology downstream of a drug is not a fusion-driven dependency

• ← bindsSETD1A — a leukemia dependency whose enzyme activity turns out to be dispensable

  WIN motif 1492-1497; counter-evidence notes MLL3 gains activity when WDR5 is absent, a built-in compensation route

• ← related toDPY30 — the smallest COMPASS subunit, and the only one that behaves like a selective dependency

  same WRAD module; WDR5 is pan-essential and uncorrelated with MEN1

• refutes →Menin is the best handle on KMT2A::AFF1 because of where the break falls, not because menin is biologically special

  MM-401 blocks the MLL1-WDR5 interaction, a contact as far past the breakpoint as KMT2A contacts go, and kills KMT2A-rearranged cells without toxicity to normal marrow

• ← part ofKAT8 (MOF) — a genuine KMT2A partner that binds the half of KMT2A the fusion discards

  the MLL1-MOF complex was purified through tagged WDR5

• ← refutesFirst systematic enumeration of the KMT2A fusion interactome, and the diff against this graph

  reported discrepancy: Yokoyama 2004 shows WDR5/RBBP5/ASH2L do not coprecipitate with fusion MLL

• ← bindsKANSL1 - the NSL scaffold, and the protein through which WDR5 binding to NSL and to MLL/COMPASS is mutually exclusive

  KANSL1 recruits WDR5 via a conserved linear motif; this interaction is mutually exclusive with WDR5's binding to MLL/COMPASS complexes.

• ← bindsKANSL2 - the second NSL subunit in the WDR5 structure, with one leukaemia record and a dependency profile that is the mirror image of a selective target

  KANSL2 binds WDR5 through a conserved linear motif; solved in PDB 4CY2.

• ← bindsNSL complex - the MOF assembly that five NUP98 fusions bind alongside MLL1, and that competes with MLL/COMPASS for WDR5

  WDR5 is required for NSL assembly and promoter targeting, and its NSL and MLL/COMPASS interactions are mutually exclusive.

• ← bindsSETD1B - the COMPASS paralogue round 4 deferred for want of a leukaemia paper, which had already been published: its SET domain sets H3K4me3 breadth and MYC in MLL-rearranged AML

  SETD1B carries its own WIN motif at residues 1745-1750, so it competes for the WDR5 pocket the graph's WIN-site dossier targets.

• ← related toWDR5 is partitioned, not shared: NSL and SET1/MLL complexes compete for one pool, so perturbing NSL changes how much WDR5 reaches the wild-type KMT2A complex the fusion depends on

  The hub whose allocation the hypothesis is about; its flat dependency profile is what a limiting shared hub would look like under either world.

• ← bindsWDR5 WIN site and KMT2A Arg3765: measured at 1.7 uM, three co-crystals, and absent from every leukaemic fusion - plus the MLL3 peptide that binds it 51-fold tighter

  MLL1 WIN peptide in the WDR5 arginine channel, Kd 1.7 uM

• ← bindsWDR5 and RBBP5: the second arm of the WRAD hub, RBBP5 V375-D376-V377 into the same propeller face, co-crystallised with the MLL1 WIN peptide in one structure

  RBBP5 V375-D376-V377 on the WDR5 propeller

• ← recruitsCARM1 / PRMT4 (Q86X55) - the H3R17me2a writer, and the enzyme whose product is read by the KMT2A C-terminus that every fusion deletes

  methyl-Pax7 bound to the MLL1/2 C-terminus recruits the ASH2L:MLL1/2:WDR5:RBBP5 H3K4 complex - so CARM1 sits upstream of WRAD by this route, in mouse satellite cells

• ← related toScoring complexes instead of genes, and the controls disagree with each other (run)

  WIN-site inhibitors act via ribosome protein gene suppression, and the ribosome scores selectively required here; convergence, not evidence

• ← related to18% of this graph's evidence is the same finding said again

  One of the four WRAD subunit entries where complex-level evidence is copied per subunit.

• binds →KMT2A / MLL1 — the gene that breaks

  binds the WIN motif, which sits in the half of KMT2A the fusion throws away

• ← refutes49 of 91 proteins in this graph have never had a drug-like molecule bound to them

  WDR5 has 63 distinct ligands, MORE than menin, and is lost from the fusion at the breakpoint - ligandability and relevance come apart, so a ligand count cannot be used to choose a target on its own.

• ← bindsKMT2B (MLL2/MLL4) - the other protein menin binds, and the one this graph never wrote

  KMT2B carries its own WIN motif competing for the same WDR5 pocket as KMT2A

• ← related toWDR5 and KMT2A are at the centrosome and the midbody - a non-chromatin address for both, with a patient-cell phenotype

  WDR5 has centrosomal and midbody addresses that UniProt does not record

• ← bindsThe reciprocal MLLT1::KMT2A transcript was reported in the paper that discovered KMT2A, and nobody has gone back to it in thirty-four years

  predicted: the reciprocal product retains the WIN motif (KMT2A 3762-3767) that docks WDR5

• ← bindsNUP98::KMT2A, inv(11)(p15q23) — the NUP98 fusion that contains KMT2A and is menin-inhibitor RESISTANT

  retains the WIN motif (3762-3767) and an intact SET domain

• ← bindsFive KMT2A-free NUP98 fusions bind MLL1 and the NSL complex — and one unbiased mass-spectrometry study failed to reproduce it

  WDR5 recovered in the NUP98-fusion pull-down

• ← related toThe standing priority asks for localization on every interactor; the graph has it on 14% of genes and 24% of its most central ones, and DOT1L has none at all

  DEGREE 39, NO LOCALIZATION: its only nuclear-compartment mentions are nucleolar stress as a drug phenotype.

• ← related toWild-type KMT2A is one of the most promiscuously labelled proteins in the proximity record - 25 of 50 baits - and menin is one of the least, at 3

  WRAD does not recover as a block: WDR5 11, ASH2L 9, RBBP5 0, DPY30 0

• ← related toRe-adjudicating the standing weak spot: proximity labelling HAS sampled AFF1, KMT2A and five leukaemia fusions - just never a KMT2A fusion

  a proximity map would not adjudicate the WRAD negative: WDR5 11, ASH2L 9, RBBP5 0, DPY30 0

• ← regulatesPRMT5 - the type II arginine methyltransferase this graph has reached for sixteen times, whose mark on H3R2 RECRUITS WDR5 to chromatin

  H3R2me2s ENHANCES WDR5 binding to the H3 tail; H3R2me2a impedes it

• ← inhibitsPRMT6 - the enzyme whose mark is MUTUALLY EXCLUSIVE with the MLL complex on the H3 tail, and which shuts HOX genes down by evicting WDR5

  H3R2me2a blocks WDR5 binding to the H3 tail and lowers MLL1/WDR5 recruitment

• ← inhibitsH3R2me2a - the asymmetric arginine mark two residues from H3K4 that LOCKS the WRAD module off the histone tail, and the first arginine mark node in this graph

  the ONLY non-permissive H3R2 state for WDR5 binding to the H3 tail; me0/me1/me2s are equally permissive

• ← bindsH3R2me2s - the symmetric mark on the SAME residue as H3R2me2a, which permits rather than enhances WRAD access, and whose 'enhancing' half this graph inherited from the paper that was later revised

  PERMISSIVE, not enhancing: equally permissive with me0 and me1 on quantitative binding plus crystallography (2020), revising the 2012 'enhances' claim

• ← regulatesArginine methylation gates KMT2-complex access to chromatin at two residues, in both directions, independently of the menin tether AND of the WIN motif - a parallel control point, not a rival to menin dependence

  the H3R2 arm gates WDR5's grip on the H3 tail, with the WIN motif and menin untouched

• ← related toIKZF3 / Aiolos (Q9UKT9) - the B-lymphoid transcription factor that goes UP under both inhibitors, regulates BCL2, and is a degradable drug target

  a cereblon-recruiting dual WDR5/Ikaros PROTAC degrades WDR5 with IKZF1 and IKZF3, dissociating the MLL/KMT2A complex from chromatin with decreased H3K4me2, and is superior to WDR5 degradation alone in MLL-rearranged leukaemia (PMID 35525905)

• ← related toRe-bound at full text, where methods live: 20 records pair a WRAD subunit with a proximity-labelling method and a KMT2A term, 14 pair one with a tiling-CRISPR term, and none is a proximity, degron or tiling study of WRAD in a KMT2A-fusion line

  no proximity, degron or tiling-CRISPR study of this subunit in a KMT2A-fusion line at full text

• ← related toThe WRAD degron absence survives, and the reason is worth more than the verdict: a 3.7 nM chemical degron of WDR5 with 2-hour kinetics already exists IN MV4;11, and it still cannot see the fusion

  adds MS67 DC50 3.7 nM, Dmax 94%, 2 h kinetics and 48-72 h washout in MV4;11, none of which gene-wdr5 records

• ← bindsKMT2A-AKAP8 (AKAP95): graded 'prey-only' by this project, and it is a direct positive modulator of MLL H3K4 methyltransferase activity from a dedicated MLL paper - ADMITTED

  WDR5 IP co-precipitates AKAP95 and tagged AKAP95 co-precipitates WDR5

• ← refutesWDR5 maintains a differentiation block — in CEBPA p30 AML, not in KMT2A-rearranged disease, and a title check nearly buried it

  blocking the MLL1-WDR5 INTERFACE with MM-401 is MLL-selective - cell-cycle arrest, apoptosis and myeloid differentiation, sparing normal marrow and non-MLL cells (PMID 24389101) - where deleting the WDR5 GENE is pan-essential, because the scaffold is shared by all six SET1-family complexes

• ← supportsUTP18 is not a KMT2A partner - it is WDR5, mis-curated into IntAct from the founding paper, and it manufactured the only KMT2A-bait gap in the interactome recount

  WDR5 is the protein PMID 15199122 actually reports in the slot IntAct fills with UTP18

• part of →KMT2A / MLL1 — the gene that breaks

  WDR5 is one of the eight polypeptides resolved in the founding MLL complex purification (PMID 15199122)

• ← related toMCM2 (P49736) - replicative helicase subunit and parental-histone chaperone in the pan-MLL-fusion core, and a live gene_exact collision with MCM7

  MLL-WDR5-RBBP5 complexes co-localise with ORC and MCM2-7 at replication origins; WDR5 or RBBP5 knockdown suppresses MCM2-7 recruitment (wild-type MLL, re-replication system)

• ← related toWDR18 (Q9BV38) - the Rix1/PELP1-TEX10-WDR18 60S maturation subunit, in five MLL-fusion interactomes but NOT the index fusion's, and the block's only member whose nucleolar partitioning is SUMO-controlled

  naming hazard only, recorded deliberately: WDR5, WDR18 and UTP18/WDR50 are three WD-repeat proteins in one literature, and a WDR5/UTP18 substitution is a documented IntAct curation error here

• ← related toUTP18 / WDR50 (Q9Y5J1) - the SSU processome subunit that recruits the exosome helicase MTREX to rRNA fragments; in three MLL-fusion interactomes but not the index fusion's, and separately the subject of a 21-year-old IntAct curation error in this graph

  documented name collision only: UTP18's synonym WDR50 / 'WD repeat-containing protein 50' against WDR5 / 'protein 5' - the substitution that caused a 21-year IntAct error here

• ← analogous toELP2 (STATIP1/StIP1) - the double beta-propeller of the Elongator core, whose informal name collides with a different human gene and whose STAT3 function is a by-similarity annotation

  same fold class - seven-bladed WD40 beta-propeller - but ELP2 has no reported histone or methyl-arginine binding, where WDR5 is this graph canonical reader and the KMT2A WIN-motif receptor

• ← bindsAFF1::KMT2A (AF4-MLL), the der(4) product: what it is made of, residue by residue

  Retains the WIN motif at 3762-3767 (Arg3765), the WDR5-binding element absent from the direct product.

• refutes →Lost from the KMT2A half: the WIN motif and the whole WRAD interface (aa 3762-3767, Arg3765)

  [contested_by] WIN-site inhibitors act in MLL-rearranged cells through ribosome protein gene suppression and p53, not an MLL1 programme

• ← bindsLost from the KMT2A half: the WIN motif and the whole WRAD interface (aa 3762-3767, Arg3765)

  Arg3765 occupies the WDR5 arginine pocket; absent from every fusion

• ← related toBET bromodomain inhibition in KMT2A-rearranged leukemia — spectacular preclinically, stalled clinically

  WDR5 WIN-site inhibitors are synergistic with BET inhibitors in MLL-rearranged cells

• ← related toThe founding WRAD negative was never run on a fusion protein

  Evidence-class correction to this entry's E_yokoyama2004 block: the WRAD measurement in that paper is on a deletion mutant, not on a fusion.

• ← related toWDR5: an established wild-type KMT2A contact, and no fusion contact ever measured

  Splits this subunit's evidence into what is established for wild-type KMT2A and what, if anything, is established for a fusion.

• ← related toFour ways a WDR5 result can be true and have nothing to do with KMT2A

  A four-point checklist - which surface, which readout, which modality, which comparator - for reading any WDR5 result as evidence about KMT2A fusions.